Ubiquitous Emotional Exhaustion in School Principals: Stable Trait, Enduring Autoregressive Trend, or Occasion-Specific State?

Abstract

Principal strain and burnout is a major issue in desperate need of further investigation and solutions. Deepening our understanding of emotional exhaustion, the central dimension of burnout, would greatly further this pursuit. Using a large, longitudinal, representative sample of Australian school principals, the present study decomposed emotional exhaustion into occasion-specific state, enduring autoregressive, and stable trait components using the STARTS (stable trait, autoregressive trait, and state) model. The results showed evidence for variance in all three components, indicating that principals’ emotional exhaustion is approximately evenly split between the enduring autoregressive component and stable trait component, with slightly less variance being observed for the occasion-specific state. Heterogeneity in this profile was mainly associated with individual characteristics of the principals themselves (i.e., experience and gender) rather than characteristics of the job (school sector and level). The results revealed that less experienced and male principals have more malleable (enduring autoregressive and state-like) emotional exhaustion, while more experienced and female principals have more trait-like emotional exhaustion. This emphasizes a likely development of emotional exhaustion from acute to chronic under persistent exposure to burnoutinducing situations, with additional evidence for a possible dispositional tendency toward emotional exhaustion. Thus, measures to tackle emotional exhaustion need to be based on the type of emotional exhaustion the principal is experiencing and ideally include elements that target both the situational/contextual and the individual factors that cause emotional exhaustion in school principals.

Full text

© 2021, American Psychological Association. This paper is not the copy of record and may not
exactly replicate the final, authoritative version of the article. Please do not copy or cite without
authors' permission. The final article will be available, upon publication, via
ttps://doi.org/10.1037/edu0000582
Online First Publication, October 7, 2021
EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
Ubiquitous Emotional Exhaustion in School Principals: Stable Trait, Enduring
Autoregressive Trend, or Occasion Specific State?
Abstract
Principal strain and burnout is a major issue in desperate need of further investigation and
solutions. Deepening our understanding of emotional exhaustion, the central dimension of
burnout, would greatly further this pursuit. Using a large, longitudinal, representative
sample of Australian school principals, the present study decomposed emotional exhaustion
into occasion specific state, enduring autoregressive, and stable trait components using the
STARTS (Stable Trait, Auto Regressive Trait, and State) model. The results showed
evidence for variance in all three components, indicating that principals’ emotional
exhaustion is approximately evenly split between the enduring autoregressive component
and stable trait component, with slightly less variance being observed for the occasion
specific state. Heterogeneity in this profile was mainly associated with individual
characteristics of the principals themselves (i.e., experience and gender) rather than
characteristics of the job (school sector and level). The results revealed that less experienced
and male principals have more malleable (enduring autoregressive and state-like) emotional
exhaustion while more experienced and female principals have more trait-like emotional
exhaustion. This emphasizes a likely development of emotional exhaustion from acute to
chronic under persistent exposure to burnout-inducing situations, with additional evidence
for a possible dispositional tendency towards emotional exhaustion. Thus, measures to
tackle emotional exhaustion need to be based on the type of emotional exhaustion the
principal is experiencing and ideally include elements that target both the
situational/contextual and the individual factors that cause emotional exhaustion in school
principals.
Keywords: Emotional exhaustion, STARTS model, School principals, State, Trait
Educational Impact and Implications Statement
Principal emotional exhaustion is a major educational concern. To help inform interventions
aimed at mitigating this concern, we examined how emotional exhaustion manifests itself in
school principals. Results from a large representative sample of Australian school principals
revealed there to be three aspects of emotional exhaustion: an acute/occasion specific state, a
chronic/stable trait, as well as an enduring component that is slow to change but not fixed.
Less experienced and male principals had more state-like emotional exhaustion, while more
experienced and female principals had more trait-like emotional exhaustion. These findings
show how emotional exhaustion develops from an acute to a chronic issue when confronted
with persistently stressful circumstances, and how female principals are more inclined to be

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
1
emotionally exhausted. We need interventions tailored towards the type of emotional
exhaustion the principal is experiencing.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
2
Ubiquitous Emotional Exhaustion in School Principals: Stable Trait, Enduring
Autoregressive Trend, or Occasion Specific State?
School principals increasingly report high levels of strain, resulting in a shortage of
qualified school leaders (Darmody & Smyth, 2016; Dewa et al., 2009; Grissom et al., 2015;
Riley, 2018; Riley et al., 2019). Indeed, changes in the educational system in response to
globalization, new technologies, and changes in workforce demographics have led to a
repositioning of the role of school principals (Dewa et al., 2009) who are already challenged
by a very diverse skill-set needed to successfully fulfil their leadership role. School principals
are required to be visionaries and direction givers, people developers, organization designers,
and teaching and learning program managers (Dadaczynski & Paulus, 2015; Liebowitz &
Porter, 2019). These role changes have resulted in more responsibility, particularly in
managerial tasks (Green et al., 2001), higher time pressure (Grissom et al., 2015), and reduced
autonomy (Riley, 2018). At the same time, principals have less resources to deal with these
increased demands, leading to the high levels of strain and attrition and consequently a lack of
qualified principals (Riley, 2015). It is not surprising that principal strain and burnout is a
major issue in desperate need of solutions to improve our school leaders' health and wellbeing
(Dicke, Marsh, et al., 2018; Wells & Klocko, 2018). In our research we investigate the
popular and well-researched emotional exhaustion dimension of the burnout construct
(Bakker & Costa, 2014), which we will show can be regarded as the central component of
burnout (Cropanzano et al., 2003; Christina Maslach et al., 2001). Although emotional
exhaustion has been the focus of a large number of studies on educator burnout (for an
overview see e.g., Dicke, Stebner, et al., 2018; Klusmann et al., 2008), there is a need for
further research. More specifically, leaders in the field of burnout have been inconsistent in
whether to describe it as a more trait- or state-like characteristic, as we will show below. In
particular, there is a need to understand if principal emotional exhaustion is an acute, a slowly

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
3
changing, or a chronic underlying issue, or even all three, and the degree to which these
components (see below for a more detailed description) differentially define the emotional
exhaustion of different types of principals (e.g., less or more experienced, male or female,
primary or secondary school etc.). Adequate description provides an avenue to more targeted
interventions. For example, should emotional exhaustion have large state and enduring
elements, event-based time-critical interventions may be essential. The current study uses
eight waves of data from a representative sample of from N = 5,509 Australian principals to
decompose principal emotional exhaustion into occasion specific state, slowly changing or
enduring, and stable trait components. We further consider heterogeneity in this profile by
examining critical demographic characteristics of gender, school sector (independent vs.
public schools), school level (primary vs. secondary), and level of experience. Overall, the
present study investigates the wellbeing of an at-risk occupational group, that is school
principals, by decomposing their emotional exhaustion into an acute, relatively enduring, and
chronic component and possible individual differences in the size of these components.
Why Principal Burnout Matters for Students Success
Increasing levels of principal burnout and emotional exhaustion is not only alarming
in itself, but also because effective leadership is crucial to a successful school environment
that fosters students’ learning (Day, 2011; Leithwood & Seashore-Louis, 2011). School
principals recognise, promote, and build the leadership capacity of staff, students, parents, and
the community, and research has demonstrated the importance of school principals for
teachers’ well-being (Dicke et al., 2019; Dicke, Stebner, et al., 2018). In turn, teacher well-
being is related to student achievement (Klusmann et al., 2016) and motivation (Dicke et al.,
2019; Shen et al., 2015). Research indicates a relation between principals' behaviours and
students’ well-being (Sebastian & Allensworth, 2012), which in turn impacts student
outcomes, such as achievement (Darmody & Smyth, 2016; Dicke et al., 2019). Indeed,

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
4
principals are the second most important influence on student learning outcomes, after
teachers (Day, 2011; Leithwood & Seashore-Louis, 2011). Moreover, some studies have
found indirect relations between principal leadership styles and behaviors, via school
environment indicators or teacher variables (Leithwood & Seashore-Louis, 2011;Sebastian &
Allensworth, 2012). These findings are supported by two recent meta-analyses that both
found strong empirical evidence for a direct relation between principal leadership and
behaviors, and student achievement (Liebowitz and Porter 2019; Wu 2020). Other studies
have also found a direct relation between principal wellbeing, that is job satisfaction, and
student achievement (Dicke et al. 2019). Thus, emotional exhaustion, which is so prevalent in
principals (Dicke, Marsh, et al., 2018; Wells & Klocko, 2018; Riley, 2018), is also likely to
significantly affect student learning. As principals’ wellbeing is related to teacher and student
outcomes, this paper provides insight into how best to approach principals’ emotional
exhaustion in the hope of not only improving their occupational health, but also providing a
means to improve the general school climate for teachers and students (Liebowitz & Porter,
2019).
State vs Trait vs Enduring
The concepts of states and traits, as well as the distinctions between them, have been
the subject of debate between researchers for several decades (e.g., Cattell, 1966; Eysenck,
1983; Spielberger, 1972). Some researchers have argued for and presented differential criteria
for the distinction of states and traits (Fridhandler, 1986; for an overview see also Hamaker et
al. 2007), some of which we will cover in the following: The most well known criterion is
temporal duration, where states are assumed to be of short duration, while traits are defined as
highly stable conditions, even life-long (Cattell, 1966). Another important criterion is
situational vs. personality causality. Here a state is assumed to be caused by the situation,
while the trait is caused by factors that lie within a person. Furthermore, Fridhandler (1986)

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
5
suggests a criterion called “concrete versus abstract entities”, which refers to states having to
be observable, or immediately available to introspection (i.e., concrete). Traits on the other
hand are abstract and thus, hard to detect, feel, or observe. Put simply, states can be defined as
a momentary condition that someone or something experiences at a specific time, for example
as a single event, such as a critical incident triggering emotional exhaustion symptoms. Traits
on the other hand can loosely be defined as a person specific enduring characteristic, which in
our case would reflect an underlying disposition for higher levels of emotional exhaustion,
such as a genetic predisposition or a chronic aspect of the person unchanged by the job
interaction. This does not entail that traits are biological. In fact, Kenny and Zautra (2001)
emphasize that trait variance could be the result of a stable environment x person interaction.
Others however, have claimed the distinction of states and traits to be arbitrary (Allen &
Potkay, 1981). Importantly, Hertzog and Nesselroade (1987) suggested that most
psychological attributes will neither be, strictly speaking, traits or states. That is, individuals’
attributes can have both trait and state components (for an extended review see also Anusic &
Schimmack, 2016). Moreover, resulting from recent statistical developments (Kenny &
Zautra, 2001; Steyer et al., 2015), it now seems clear that individual attributes can also consist
of a third component which lies in-between state and trait; the autoregressive trait. This
autoregressive trait component is also referred to as enduring (Wagner et al., 2016), and we
will use the term enduring autoregressive component hereafter. This is because it is neither
constant like a trait nor transitory like a state, and for the present study it could be translated
as the development of job-related emotional exhaustion over time. Jansen et al. (2020)
describe the component as variance that “refers to individual differences that can be attributed
to influences that change over time (i.e., they do not show the same effect across all
measurement points) but are still partly stable (i.e., they endure across several measurement
points)”. With regard to the aforementioned criteria , this autoregressive component is

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
6
enduring, but can change slowly over time (Mund et al., 2020), and can even disappear over
time. It would most likely result as an interaction of the situation and person, but not
immediately, rather as a result of ongoing stimulus and a disposition to react to that stimulus.
It is concrete, at least in the beginning, but might transition into becoming an abstract entity as
the person becomes “used to it”. For example, a policy change may lead to changes in the
principal’s job that have ongoing effects on emotional exhaustion over many years. Enduring
aspects of emotional exhaustion are often an inherent part of the job where, for example,
large-scale educational policy changes tend to be aligned to multi-year election cycles. In the
beginning, the effects of the policy change cause several isolated stress reactions, over time
this continuous exposure changes into an enduring or lingering experience of strain. Another
change to policy might, however, change or even nullify that experience. To summarize, there
is still an ongoing discussion on the distinction of states and traits, which has recently been
enriched by the proposal of a third slowly changing, but malleable component, which lies in-
between state and trait.
Burnout, Emotional Exhaustion and their Manifestation as State or Trait, or Enduring
Phenomenon
To date, research on burnout, and thus emotional exhaustion, has found evidence for
burnout appearing as state-like, trait-like, or both, depending on individual differences.
According to the manual of the most used instrument to assess burnout in the workplace—the
Maslach Burnout inventory (Maslach et al., 1996)—burnout has been defined as a malleable
state that manifests itself as exhaustion, with cynical and self-doubting experiences. One year
later, the Maslach Burnout inventory (Maslach et al., 1997) updated their terminology to
“enduring state” (rather than a state) which suggests that any change would be slow. Shortly
after, the most popular definition of burnout was coined: “Burnout is a prolonged response to

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
7
chronic emotional and interpersonal stressors on the job, and is defined by the three
dimensions of exhaustion, cynicism, and inefficacy” (Maslach et al., 2001; p.397). This
defines burnout as something that seems to be ongoing or continuous, so more trait-like, and
consists of three dimensions: emotional exhaustion, meaning feelings of being emotionally
drained and fatigued; depersonalization, meaning a callous or cynical attitude—in this case,
mostly towards the teacher’s students; and reduced personal accomplishment, meaning a
person’s negative evaluation of their own abilities and achievements.
In the early years of burnout research there was another emerging conceptualization of
burnout by Golembiewski and colleagues (1985; 1988) that agreed with the Maslach model.
This new conceptualization defined the same three dimensions, but proposed a different
process. Further, this model also differentiated between acute and chronic burnout, which
could resemble a state and trait-like distinction (Golembiewski, 1985; 1988). It is this model
by Golembiewski that is the motivation for this research in decomposing burnout into state,
enduring, and trait components.
Regarding the distinction between acute and chronic burnout, Golembiewski et al.
(1985; 1988) argue that acute burnout is a result of a sudden severe exposure to stress and
thus, more likely to be a result of personal trauma, while chronic burnout is a steady gradual
process that arises through long-term exposure to work stress. Golembiewski et al. (1985;
1988) claim that their model represents chronic burnout, while the Malsach Model represents
acute burnout (see Lee & Ashford, 1993). Indeed, in most recent publications, burnout is
repeatedly referred to as “state” (e.g., Maslach & Leiter, 2016). Bakker and Costa (2014),
however, show that burnout can last over long periods of time and, surprisingly, deter the use
of the term chronic burnout, which would imply that there is also acute burnout similar to the
assumptions of Golembiewski et al. (1985; 1988).
In the present study we will focus on emotional exhaustion which “is the central quality

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
8
of burnout, the most obvious manifestation of this complex syndrome, the most widely
reported, and the most thoroughly analyzed.” (Pogere et al., 2019, p.270). This is particularly
true in research on educator wellbeing (for an overview see Arens & Morin, 2016).
Cropanzano et al. (2003) and Maslach et al. (2001) both suggest that emotional exhaustion
has stronger relations to important outcome variables than the other dimensions of burnout
(see also Lee & Ashforth, 1993; Schaufeli, 1998). This evidence is supported by recent
research linking biological responses to psychological phenomena which demonstrates the
significant role of emotional exhaustion in relations with psychophysiological variables (for
an overview, see Kanthak et al., 2017). In addition, conceptually, Cropanzano et al. (2003)
reported the work of Schaufeli and Enzmann (1998), who found that individuals who state
they are “burnt out” are mostly referring to feelings of emotional exhaustion. Overall, the
development of burnout and particularly the manifestation of its central component emotional
exhaustion as more state, or trait-like are still unclear and, thus, further research is much
needed.
Why is it important to decompose Emotional Exhaustion into State, Trait and Enduring
Variance?
Researchers have been increasingly calling for more research on the relative
contributions of state and trait measures on organizational outcomes for some time now (e.g.,
Simbula, 2010; Sonnentag, 2005; Wright et al., 2003). Investigating the decomposition of
emotional exhaustion into state and trait-like components is important for three reasons:
theoretical reasons; empirical reasons; and practical reasons. Ultimately, the most important
practical goal is to develop a sufficient understanding of emotional exhaustion in order to
prevent its development and to inform its treatments. A strong theoretical foundation and
appropriate empirical and methodological approaches are important prerequisites for adequate
practical implications and translation of research into practice. Hence, in the following, we

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
9
will describe how the present study will contribute to all three aspects: theory, empirical
measurement of emotional exhaustion, and finally practical implications.
Theoretically, although our study focuses on emotional exhaustion, we will also be able
to derive important conclusions to refine the overall definition of burnout. One aspect of this
is to clarify the temporal nature of emotional exhaustion (and thereby burnout) as an acute
state-like, or a chronic trait-like syndrome, or a mix of both that would appear as an enduring
slowly changing (autoregressive) concept as discussed above. Related to this, we collect
evidence for internal (trait) or external (state) causation. This would be of great value for
further developing the conceptual models dealing with the antecedents of emotional
exhaustion. The most popular and recent models (Maslach & Leiter, 2016) in this regard are
the Job-Demand Resources (JD-R) model (Bakker & Demerouti, 2014) and the Conservation
of Resources (COR) Model (Hobfoll, 2001). Both models assume that burnout is a result of an
imbalance of demands and resources. More precisely, in case of the JD-R model, burnout
develops as a result of too many demands that lead to strain, the so-called health impairment
process, and too few resources to deal with these and reduce them. The COR model proclaims
that burnout develops due to threat and loss of resources. Both of these models make the most
sense under the assumption that burnout is more state-like and caused by a situational causal
context of resource-demand imbalance (Bakker & Costa, 2014). However, Maslach and Leiter
(2016) also state that the factors causing burnout can be both situational and individual (see
also Wright et al., 2003). Indeed, the JD-R model explicitly includes personal and job
resources. These theoretical insights are critical for developing appropriate and targeted
interventions to prevent and/or treat high levels of emotional exhaustion as needed (see below
for more details).
Empirically, there has been some discussion on how to measure and empirically model
states or traits. For assessing a state, it is suggested that items should ask for something the

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
10
participant feels right now or today, while traits are reflected by items that ask how a
participant feels about something in general. Moreover, researchers suggest that a trait cannot
be assessed by a single occasion or one-time measurement but rather by an aggregation of
several measurement occasions (Fridhandler, 1986; Steyer et al., 2012). This is because traits
are very complex and abstract (see above distinction for states and traits) and are never “here
and now”. Measuring the occurrence of a phenomenon once, does not allow for an inference
of whether a construct is state or trait. Technically, variation in both trait and state may
contribute to construct variation, which is confounded and cannot be sufficiently teased apart
based on one-time measurement (Hamaker et al., 2007; Hamaker et al., 2015).
There are, however, an increasing number of researchers that have also tested models,
mostly using diary studies or experience sampling, that use very frequent measures (i.e.,
daily) of occupational well-being, including emotional exhaustion (e.g., Aldrup et al., 2017;
Klusmann et al., 2020; Simbula, 2010). These models focus on the within-person
development of emotional exhaustion over time. Putting these two research traditions, that is
focussing on traits or states, together, along with the development of more advanced statistical
models, allows us to model both trait and state simultaneously. Interestingly, most studies
investigating emotional exhaustion so far have operationalized emotional exhaustion as a
state-trait hybrid like construct which is repeatedly measured over time, but with long time
periods between measurement points (Bakker et al., 2014; Bakker & Demerouti, 2007; Dicke,
Elling, et al., 2015; Dicke, Stebner, et al., 2018) and primarily focusing on between-person
differences (Simbula, 2010; Sonnentag, 2005). Thus, in most research emotional exhaustion is
neither constant like a trait nor transitory like a state and instead reflects slow change over
time which matches the aforementioned definition of burnout as an “enduring state” (Maslach
et al., 1997). In the present study, we will also measure a variance component that reflects this
slow change over time, namely an enduring autoregressive component (see Wagner et al.,

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
11
2016; see also Method section for more details). The adequate measurement and identification
of the variance components of emotional exhaustion is also immediately important for
practical implications, to appropriately diagnose and identify the needs of individuals
experiencing emotional exhaustion. Further, focusing on the within person development of
emotional exhaustion instead of between person differences gives valuable insights for the
effectiveness of measures targeting emotional exhaustion.
Practically, preventing and treating emotional exhaustion, its symptoms, and the
consequences thereof would differ largely depending on it being an occasion-specific or
momentary condition (state), or a person-specific enduring characteristic (trait). Likewise,
exploration of profiles of variance for different types of principals may help develop more
targeted interventions. If emotional exhaustion is predominantly state-based due to an
overload of demands (Huang et al., 2011) it could be resolved by short-term investment of
resources to address acute demand/resource imbalances (Bakker & Demerouti, 2014). For
principals this could mean a decrease in the sheer quantity of work, increasing job specific
self-efficacy, and/or increasing social support of colleagues (Dicke, Marsh, et al., 2018; Dicke
et al., 2019). If burnout is predominantly chronic or trait-like, clinical interventions which aim
to change persistent inherent characteristics of the individual may be more appropriate (Steyer
et al., 2015). Enduring aspects may require both timely investment of resources as well as
longer-term personal interventions to manage the legacy of acute events. Thus, further
investigating the experience of emotional exhaustion as more state, or trait-like, or anything
in-between would be of major benefit for theoretical clarity and adequate empirical
measurement of the construct (and possibly of all its components). Most importantly, more
information on the nature of how emotional exhaustion manifests itself is needed for deriving
effective measures that could prevent or treat any experience of emotional exhaustion.
Hypothesis and Research Questions

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
12
In the present study we will investigate the composition of school principals emotional
exhaustion by modelling a series of STARTS (Stable Trait, Auto Regressive Trait, and State;
Kenny & Zautra, 2001; Kenny & Zautra, 1995) models using data from a large, longitudinal
and representative sample of Australian school principals. The STARTS model enables
researchers to model three different sources of variance in repeated measures: 1) a stable trait
component which is assumed to be time-invariant (ST); 2) an auto-regressive component,
which is assumed to be time-varying (ART); and 3) a state component which is assumed to be
completely occasion-specific (S; no stability over time; Kenny & Zautra, 2001). Wagner et
al., (2016) describe this enduring autoregressive component as a time-varying factor that
depends on the previous time-point and a random component (see Method section for more
details). Our review of burnout theory and empirical studies on emotional exhaustion
indicated that emotional exhaustion should consist of all three variance components; variance
that can be assigned to 1) a time-invariant stable trait component, i.e., an underlying
disposition to emotional exhaustion; 2) a time-varying, but enduring autoregressive
component i.e., the development of job related emotional exhaustion over time; and 3) an
occasion-specific state component, i.e., as single event, such as a critical incident triggering
emotional exhaustion symptoms (Kenny & Zautra, 2001). However, so far most research has
focussed on emotional exhaustion as an enduring autoregressive construct (Bakker et al.,
2014). This leads to our first hypothesis:
Hypothesis 1 (H1) : We expect emotional exhaustion to show variance attributable to
all three components, that is stable trait, enduring autoregressive component, and occasion
specific state.
Our second group of hypotheses/research questions are based on the inconsistent

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
13
assumption and operationalisation of emotional exhaustion as state, enduring, or trait-like, and
are about how heterogeneity in these components of emotional exhaustion manifests itself.
Such heterogeneity may be associated with important individual characteristics. In the present
study we investigated potential differences in the sizes of occupation specific state, enduring
component, and stable trait due to experience, gender, school level, and school sector.
Identifying such differences is important for identifying potential personal or
occupational risk factors associated with all three variance components of emotional
exhaustion. If for example a certain school sector is more prone to develop more trait-like
emotional exhaustion, then interventions for that sector should be targeted towards treatment
of a more chronic manifestation of emotional exhaustion (see discussion section for more
details).
Regarding effects of job experience there is evidence that the stability of
characteristics might change with maturation (age and/or experience). The leading research in
this area is based on personality traits (Prenoveau et al., 2011), followed by research in
relation to depression. In the seminal work on personality traits by (Costa & McCrae, 1988) it
was assumed that personality traits change until individuals reach the age of 30, but remain
stable thereafter. A study that looked at personality traits as well as anxiety disorders and
depression revealed similar results (Prenoveau et al., 2011). However, this study found that
depression, which is considered to be similar to emotional exhaustion (Brenninkmeyer et al.,
2001), showed more state-like tendencies in adults than either anxiety disorders or personality
traits. Emotional exhaustion, and depression for that matter, are not personality traits and their
increasing stability might be better described by symptoms manifesting from an acute state to
a chronic disease as discussed by Bakker and Costa (2014) and Golembiewski and
Munzenrider (1988). Thus, the trait variance could be the result of a very demanding and
stable environment x person interaction. In the same manner, research shows that PTSD will

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
14
most likely develop in those most frequently involved in distressing incidents and the highest
levels of cumulative trauma and risk exposure (Milligan-Saville et al., 2018). More
experienced principals could, furthermore, be caught in a negative burnout cycle, where
already depleted resources are drained to battle ongoing demands (Bakker & Costa, 2014;
Bakker & Demerouti, 2014; Hobfoll, 2001). Applying these findings to the present study
would indicate that emotional exhaustion should show higher state variance for less
experienced principals, while it should show more enduring variance, and in particular stable
trait variance, in more experienced principals.
Regarding gender, so far there seem to be no studies that have looked at differences in
the makeup of emotional exhaustion as to its state, enduring, or trait components. Instead
research has mainly investigated gender differences in the mean levels of emotional
exhaustion, which is not directly comparable to differences in variance components, but could
give a rough indication for formulating hypotheses. Results of a large meta-analysis including
409 effect sizes from 183 studies showed that women tend to report higher levels of emotional
exhaustion than men (Purvanova & Muros, 2010). Studies focusing on principals however,
are inconclusive with some studies showing that female school principals report higher levels
of mental health-related problems including emotional exhaustion (Dadaczynski & Paulus,
2015; Weber et al., 2005) similar to the general population, while others do not find such (or
only negligible) differences (Darmody & Smyth, 2016; Dewa et al., 2009; Friedman, 2002).
Similarly, research has only examined mean differences in emotional exhaustion for
school levels and school sectors. However, this research has focused on teachers, not
principals. Further, research on mean levels of burnout and emotional exhaustion regarding
school level is inconsistent. While Klassen and Chiu (2010) presented findings that support
less teacher strain working with younger students, other researchers could not support these
findings (Antoniou et al., 2000; Dicke et al., 2016). One of the rare studies that sampled

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
15
principals did not find big differences between the burnout levels of elementary and
secondary school principals either (Friedman, 2002). Research on the effects of school type or
sector on the mean level of school principal burnout is scarce. The studies we could find
indicated no meaningful differences (Riley, 2018). Based on these findings we formulated a
hypothesis and several research questions:
Hypothesis/Research Question 2(H2):
a. For less experienced principals we expect the state component to be the largest,
followed by the enduring autoregressive component, and then a small trait
component. Simultaneously, we expect the opposite pattern for principals with
more experience in leadership, and a slightly more balanced (even distribution)
pattern for principals with five to ten years of leadership experience.
We leave as a research question if there are any differences in the sizes of the stable
trait, enduring autoregressive component, and occasion specific state components of
emotional exhaustion variance between:
b. female and male principals.
c. principals working in different school sectors.
d. principals working in different school levels.
Overall, we expect to find evidence for all three components in our sample of school
principals. We expect the trait components to increase and the malleable components to
decrease with increasing levels of experience while we leave as research questions all other
individual differences.
Method
Participants

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
16
Participants were school principals working in Australia during 2011–2018. The
sample (N = 5,509) comprised 41.7% male and 58.3% female participants: 70.3% principals,
24.5% assistant/deputy principals, 2.0% campus principals of a multi-campus school, and
3.8% other1. The mean age of school principals in our sample was 58.22 (SD = 8.17) years.
Regarding the school levels the principals managed, 58.5% were primary schools, 26.3%
were secondary schools, and 13.3% were combination schools (both primary and secondary).
Further, 74.7% of the sample worked at public schools, while 14.2% worked at catholic and
11.1% worked at independent schools. The mean years of experience (in 2011- our first wave)
in their current position was 6.44 years (range = 0-37) and 15.26 years in leadership roles
generally (range = 0-47).
Data on these school principals were collected as part of a large research project on
principal health and wellbeing (Dicke, Marsh, et al., 2018; Riley, 2014, 2017), where
principals filled in a large survey annually over eight waves in 2011–2018. The emotional
exhaustion scale was completed by a total of 5,509 principals at least once and on average by
n = 2,329 principals a year (see Table 1 for details). Other publications in peer-reviewed
journals based on this data are Beausaert et al. (2016), Dicke et al. (2018) and Maxwell and
Riley (2017).
Measures
Emotional Exhaustion
We used the emotional exhaustion scale of the Copenhagen Psychosocial
Questionnaire (COPSOQ-II) developed by a consortium of occupational health and wellbeing
researchers led by Tage S. Kristensen (Kristensen et al., 2005) as a tool for practice and
research (see Dicke et al., 2018 for an overview). The scale consisted of four Likert scaled
1 Supplemental Materials revealed no significant differences in the decomposition of emotional exhaustion for
principals vs. deputy/assistant principals. For details see Supplemental Material Table S3.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
17
items (on a five-point scale ranging on a continuum from 1- all the time to 5- not at all; for
histograms of response distribution please see Table S2 in Supplemental Material). Items
were: “How often have you felt worn out?”, “How often have you been physically
exhausted?”, “How often have you been emotionally exhausted?”, “How often have you felt
tired?”. This scale showed an average mean of M = 2.8 (see Table 1). McDonald’s (1999)
Omega, which reflects the proportion of variance in the scale scores accounted for by a
general latent factor, is reported as a measure of internal consistency (see also Zinbarg et al.,
2006). Omega coefficients were .92 on average. All time points had omega values above .91
(see Table 1 for details).
Covariates
We included several individual characteristics as grouping variables. These included
gender, average time in leadership positions (less than five years, five to ten years, more than
ten years; based on principals’ levels of experience in 2011), school sector (public vs catholic
and independent), and school level (primary vs secondary). The cut-off values for leadership
experience (up to 5 years, up to ten years, more than ten years of leadership experience) were
based on the fact that educator strain and attrition is usually highest in the first five years of
working in the profession, then shows a slight decline (reflected in our 5-10 years of
experience group) and then increases steadily again towards the end of the career (Dicke,
Parker, et al., 2015); and specifically for principals see (Goldring, 2014, 2018; Hanselman et
al., 2016). We merged the catholic and independent schools to one “non-public schools”
group (see Supplemental Material for more details on the differences and similarities of
public and non-public schools in Australia) and excluded the groups of principals responsible
for primary as well as secondary students as they were very small in size (see participants; see
Table 2).
Statistical Analysis

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
18
STARTS Model
For investigating our research questions we used the STARTS (Stable Trait, Auto
Regressive Trait, and State; Kenny & Zautra, 1995; Kenny & Zautra, 2001) model which
accounts for three different sources of variance in repeated measures: 1) a stable trait
component which is assumed to be time-invariant (ST), hence correlations of the stable trait
factors over time are all one; 2) an enduring auto-regressive component, which is assumed to
be time-varying (ART), with a correlation structure that is assumed have larger correlations
over shorter time periods and smaller correlations over longer time periods (i.e., a simplex);
and 3) a state component which is assumed to be completely occasion-specific (S), thus,
having no stability over time (Kenny & Zautra, 2001). Similarly, for coefficients in relation to
stability the stable traits coefficient is 1, the coefficient of the enduring autoregressive
component b can reach any value between 0 and 1, and the states coefficient is 0. In brief we
will investigate the following components in all our models: VST the stable trait variance
component, VART the enduring autoregressive variance component, VS the occasion specific
state variance component, and we will estimate b as a measure of the stability of the enduring
autoregressive factors. We will report the relative variances, where, for example, the occasion
specific state variance is calculated as a proportion of the sum of state, stable trait, and
enduring autoregressive components variance, such that VS/(VST + VART + VS).
It is important to emphasize that the time intervals used for STARTS models affect the
interpretation and differentiation of the STARTS components (Anusic et al., 2012). This
means that as we used a one year interval our occasion specific state component included all
influences from that year. As our overall study had a duration of eight years, the STARTS
model will thus be able to clearly differentiate between all three components (Anusic et al.,
2012). However, it also means that our occasion specific state is not necessarily comparable
to the state-like measures of burnout in other literature based on shorter time intervals, such as

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
19
those present in diary of experience momentary assessments (e.g., Aldrup et al., 2017;
Simbula, 2010). For the present study, which did not focus on the specific events or
circumstances leading to or increasing the levels of emotional exhaustion, but rather the
decomposition of emotional exhaustion, a time interval of one year was thus appropriate. It is
important to note that although the waves are measured a year apart they are still situated in a
particular time and thus the state component captures the here and now of the principal's
experience in the moment where they are filling out the survey.
In order to identify the STARTS model certain constraints are necessary. Assuming
that the waves are equally spaced, Kenny and Zautra (2001) suggest constraining the rate of
change to be the same between all pairs of adjacent waves. Further, stationarity is assumed,
which means that the overall size of the variance and the partitioning of the variance into the
three components is stable over time. In order to maintain consistency of variance partitioning
across time, the variance of the the disturbance of the first enduring autoregressive factor was
constrained to equal VART(1 − b2) (Donnellan et al., 2012; Kenny & Zautra, 1995). In cases
where there are enough measurement occasions some of these constraints can be relaxed. In
the early versions of the STARTS model, based on a single manifest variable, the latter
occasion specific state component or time specific effects were conflated with measurement
error. Thus, we modelled a latent (second order) version of the model where every time point
was based on a multiple indicator factor that explicitly accounts for measurement error
(Alessandri et al., 2016; Donnellan et al., 2012; Mund et al., 2020; Wagner et al., 2016). This
means variables in the STARTS model were purged of measurement error and none of the
reported variance is attributed to error (for an overview of items residuals of our basic
STARTS model please see Table S2 in the Supplemental Material). The latent STARTS
model is essentially a more restricted model of the (Marsh & Grayson, 1994) general
longitudinal model that includes all possible correlations between latent factors at all each

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
20
wave (see Donnellan et al., 2011 for more details; see also Kenny & Zautra, 1995).
Multigroup Models
To investigate factors that could explain differences in the variance components we
tested several multigroup models where we could compare the distribution of variance
components between groups. The advantage of using the multigroup approach over covariates
is that we will be able to simultaneously decompose the variance components of the STARTS
model for all groups (Mund et al., 2020; Wagner et al., 2016). Difference tests between
components were conducted with the DELTA method using MPlus’ model constraint
command. To ensure time invariance of our grouping variables, we excluded principals that
had changed either school sector (N= 17) or school level (N= 42).
Longitudinal models such as the STARTS models rely on the assumption that the
measure consistently functions in the same way, i.e. ranks the people in the same way at each
occasion. Hence an important prerequisite for testing our models was to establish metric
invariance in the overall sample over time and across samples in the multi-group models. In
our data, measurement invariance was tested in all models by comparing fit of models with
freely estimated factor loadings to corresponding models with factor loading constrained
across time or groups (where feasible). In all cases results revealed strong evidence for
measurement invariance (see Supplemental Material Table S1).
Missing Data
For the present analyses, we include all participants who responded in at least one of
eight years, 2011-2018. Principals are invited to participate in each wave, even if they did not
respond to the previous wave(s). It is important to note that in this sample dropout does not
follow the same pattern as in many other studies; instead of principals dropping out not
returning, most principals drop-out occasionally and return to the survey sometime in the
following years. In 2018, approximately 94.5% of participants have participated in the survey

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
21
multiple times (Riley et al., 2019). The response rates for each wave are between 34.5%-52%
with an average of 41% across all waves (total N of 5,509; for details see Table 1). As such,
we applied multiple imputation to our data. Thus, we were able to use the data of all 5,509
principals at each wave. As there are about ~10,000 principals in Australia (Riley et al.,
2019), our sample represents around half of Australian principals at each wave (with just
under a quarter having provided data at each wave). There was no systematic or meaningful
pattern of relationship between survey non-response and levels of emotional exhaustion (see
Supplemental Material for more details).
Multiple imputation provides a powerful tool for dealing with missing values by
producing valid parameter estimates that are less biased than ad hoc procedures such as
listwise deletion, even in cases where data are not missing at random (Schafer & Graham,
2002). Multiple imputation has been found to result in trustworthy, unbiased estimates for
missing values even when large numbers of values are missing (Enders, 2010) and to be an
adequate method to manage missing data in large longitudinal studies (Jelicić et al., 2009).
More specifically, as emphasized in classic discussions of missing data (e g., Newman, 2014),
under the missing-at-random (MAR) assumption that is the basis of multiple imputation,
missingness is allowed to be conditional on all variables included in the analyses, but does not
depend on the values of variables that are missing. In a longitudinal panel design, this implies
that missing values can be conditional on the values of the same variable collected in a
different wave. This makes it unlikely that MAR assumptions are seriously violated, as the
key situation of not MAR is when missingness is related to the variable itself. Hence, having
multiple waves of parallel data provides strong protection against this violation of the MAR
assumption. We also specified the missing data model assuming multivariate normality. We
used the built in Mplus data imputation command where the IMPUTE option is used to
specify the analysis variables for which missing values will be imputed. This handling of the

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
22
missing data is consistent with the STARTS model (substantive model compatible) because
the STARTS model is only based on the covariance structure and this is reflected in the
imputation model in Mplus. In effect, only variables that are part of the analyses are included
in the imputation procedure. There were no additional auxiliary variables included for the
imputation process. The reported models are based on 100 imputed datasets. The final
parameter estimates were obtained through the aggregation procedure implemented in Mplus,
following Rubin’s (1987) rules.
Multiple imputation in Mplus is carried out using Bayesian estimation. Data are
imputed using an unrestricted (saturated) model, which is the model of unrestricted means,
variances, and covariances for all continuous items of emotional exhaustion. Imputation
models were estimated with 10,000 Markov chain Monte Carlo (MCMC) iterations with two
Markov Chains (Muthén & Muthén, 1998–2017). Convergence of imputation models is
evaluated by the potential scale reduction (PSR; Asparouhov & Muthén, 2010). PSR is the
ratio of total variance across chains and pooled variance within a chain. We used PSR < 1.05
as an appropriate convergence criterion (Gelman & Rubin, 1992). Every 100th iteration in the
draws from the posterior distribution are used for imputed values.
Model Fit
For all structural equation modeling we used Mplus (Version 7; Muthén & Muthén,
2012). Given the known sensitivity of the chi-square test to minor deviations from
multivariate normality, and to minor misspecifications in large sample sizes, applied SEM
research focuses on indices that are relatively sample-size independent (Hu & Bentler, 1999;
Marsh et al., 2004). This includes the Root Mean Square Error of Approximation (RMSEA),
the Tucker-Lewis Index (TLI), and the Comparative Fit Index (CFI). Population values of TLI
and CFI vary along a 0-to-1 continuum, in which values greater than .90 and .95 typically
reflect acceptable and excellent fits to the data respectively. Values smaller than .08 and .06

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
23
for the RMSEA support acceptable and good model fits respectively (e.g., Browne & Cudeck,
1992; Hu & Bentler, 1999).
For investigating our research questions and hypotheses we present an a priori series
of sequential models:
1. The basic STARTS model of emotional exhaustion (H1)
2. Several multigroup STARTS models of emotional exhaustion, based on Model
1a, for identifying differences in the variance components based on group
membership (H2)
Results
We will first test our basic STARTS models with the entire sample before moving on
to looking at a STARTS model including covariates that might predict differences in the
variance components. Then we will test multi-group models, based on those covariates, for
investigating differences in the variance components over groups, similarly to the strategy
used by Wagner et al. (2016). As a prerequisite for our analyses, we tested longitudinal
invariance of emotional exhaustion. Results comparing model fit indices revealed that fit of
the more restricted model, where factor loading were being held invariant, was not worse with
regard to fit indices (see Supplemental Material Table S1). The model with invariant factor
loadings also provided latent correlations over time (see Table 2) with coefficients reflecting
test-retest correlations.
In our basic STARTS model we modeled all variance components: a) a stable trait
factor; b) eight enduring autoregressive factors; c) eight occasion specific state factors; and d)
measurement factors for the eight waves as latent variables (see Figure 1).
Basic STARTS Model (H1)

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
24
In this basic model we assumed stationarity2 of total variance in addition to the
proportion of variances to be invariant over time. Results revealed excellent model fit (see
Table 2). Inspection of the variance components showed that almost equal parts of the burnout
variance could be attributed to the enduring autoregressive component (39%), the stable trait
component (33%), and the remaining 27% being attributable to the occasion-specific state
variance. All of these components were statistically significant3.
The autoregressive path coefficient, which represents the 1-year stability of the
enduring autoregressive factor, was high (.86). Following Donnellan et al. (2012), we used
this estimate to calculate the connection between the enduring autoregressive components in
2011 and 2018, by raising the value of the path coefficient (.86) by the 8th (2011-2018 = 8
years) power resulting in a correlation of .29, which is moderate to small, but suggests that
some part of the test-retest correlation between the emotional exhaustion scores of 2011 and
2018 (r =.50; SE = 0.03; see Table 2) reflects a significant enduring autoregressive effect
(Donnellan et al., 2012), while the rest of this correlation is driven by the stable trait factor
(by definition the occasion specific state factor can not have shared influences).
Overall, these results suggest that, for the entire sample, principals’ emotional
exhaustion could be ascribed to enduring (autoregressive) or steady changes , but also
depended in large part on very stable characteristics as well as occasion specific conditions.
Multigroup Models (H2)
All multigroup models (Models 2a-d) were based on Model 1 but included a grouping
variable based on several time invariant characteristics that we assumed might have potential
2 We conducted additional analyses (see Supplemental Material) where we tested for quasi-stationarity and freed the
autoregressive coefficients over time. As these models revealed negligible differences to this basic model, we decided that
there was no benefit in either freeing the overall size of the variance or autoregressive coefficients over time. Thus, all
consequent models are based on the more parsimonious model assuming stationarity and invariant autoregressive
coefficients.
3 Results reported here are based on ML. Running the models with MLR revealed very similar results with differences of less
than .01 in the size of the variance components.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
25
effects on the variance components of emotional exhaustion. These variables were years in
leadership position (Model 2a; 5 years and less, five to ten years, and more than ten years of
experience in leadership positions), gender (Model 2b), school sector (Model 2c; i.e., public
vs. independent and catholic schools), and school level (Model 2d; i.e., primary or secondary
school). All models fit the data well (see Table 3). The variance components however, varied
in size depending on group memberships.
Leadership experience
Results revealed a pattern in line with our hypothesis (see Table 4 and Figure 2). Both
groups with lower levels of experience showed much larger malleable components (enduring
autoregressive and occasion specific state). For both groups, the enduring autoregressive
component was largest less than five years: 59%; five to ten years: 57%) followed by the
occasion specific state (less than five years: 31%; five to ten years: 29%). The stable trait
component for both groups with less leadership experience was smallest and, in fact, not
statistically significant (see Table 4).
For the group with more than ten years of experience the stable trait component made
up 44% of their variance in emotional exhaustion. Nevertheless, only the differences in
occasion specific state and stable trait of the group with up to five years of experience, and
over ten years of experience were statistically significant. The autoregressive path estimate
was high in all three groups, and there were no significant differences between the groups in
either the autoregressive path estimates nor the autoregressive component. Overall, for more
experienced principals, emotional exhaustion could be ascribed to very stable characteristics,
while for less experienced principals emotional exhaustion depended in a large part on
malleable conditions.
Gender
Examining gender specific patterns in the variance components of emotional

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
26
exhaustion revealed that, for women, a larger part of emotional exhaustion variance was
attributed to the stable trait component (48%, occasion specific state was 21%, autoregressive
trait was 31% ), while for men the variance attributed to autoregressive trait was highest
(45%), followed by occasion specific state variance (30%), with slightly smaller stable trait
variance (25%; see Figure 2). Only the difference in the state component was statistically
significant between women and men among three components. The autoregressive path
estimates were again high in both groups, and there were no significant differences. Overall,
for female principals, emotional exhaustion could largely be ascribed to very stable
characteristics, while for male principals emotional exhaustion depended on malleable
conditions.
School sector
Results based on school sector (public vs. non-public) showed the enduring
autoregressive trait to be largest in both groups (public 45%, and non-public 47%). Stable trait
and occasion specific state components for the public-school principals were almost the same
size (29% vs. 26%, respectively; see Figure 2) while the occasion specific state component
(32%) in non-public school principals was a bit larger than the stable trait (22%), which
additionally was not statistically significant. Autoregressive path estimates were high. There
were no statistically significant differences in any components or the path estimates between
these groups (see Table 4). Overall, there were no differences in the size of the variance
components for principals of public or non-public (private/independent) schools.
School level
The variance components for principals from different school levels (primary vs.
secondary) showed a very similar pattern, with principals of secondary schools showing very
similar sized stable trait (39%) and enduring autoregressive trait (38%) components and a
smaller occasion specific state component (see Figure 2). Principals of primary schools

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
27
showed a slightly larger enduring autoregressive trait (41%), followed by the stable trait
(31%) component and occasion specific state component (28%). The stable trait and enduring
autoregressive trait for both groups was not statistically significant. The autoregressive path
estimates were high and similar in size. There were no significant differences between groups
in any of the components or the autoregressive path estimates. Overall, there were no
differences in the size of the variance components for principals of primary or secondary
schools.
Discussion
The major aim of the present study was to investigate a decomposition of emotional
exhaustion into its occasion specific state, enduring autoregressive, and stable trait
components. Results revealed that, for principals, emotional exhaustion is approximately
evenly split between the enduring autoregressive component, stable trait component, and
occasion specific state component. The heterogeneity in profiles, with regard to the variables
included in the present study, was mainly associated with individual characteristics of the
principal themselves (i.e., gender and experience) rather than characteristics of the job (i.e.,
school sector and level). This finding is consistent with findings from Klusmann et al. (2008)
who found that school‐level characteristics, as opposed to individual teacher differences,
accounted for only a small amount of the variance in teachers’ emotional exhaustion.
Theoretical Implications for a Multi-Componential Emotional Exhaustion
The strength of using the STARTS model is that it also allows testing for an occasion
specific state, a stable trait, and a third enduring autoregressive, but still malleable component.
Our results show that emotional exhaustion has meaningful levels of all three components:
occasion-specific state, enduring autoregressive component, and stable trait. This leads to the
question of whether the components represent stages of how emotional exhaustion manifests
from an acute state to a chronic disease as suggested by other researchers (Bakker & Costa,

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
28
2014; Golembiewski & Munzenrider, 1988), and as is similar to findings in the trauma and
PTSD literature (Milligan-Saville et al., 2018). Our results suggest this is true for emotional
exhaustion, given the change in variance components favoring trait in experienced principals.
We found the biggest differences in the components resulted from the principals’ level of
experience. Here, it was unsurprising that inexperienced principals' emotional exhaustion
variance was almost completely explained by state and enduring autoregressive components,
while more experienced principals showed a larger trait component. In effect, emotional
exhaustion seems to be an ongoing process that unfolds over time in line with assumptions by
Bakker and Costa (2014). Based on the COR model this manifested emotional exhaustion is
the result of a loss spiral, where the ongoing depletion of resources leads to increasing levels
of emotional exhaustion. A possible cause for this spiral is the so-called process of
undermining (Linden et al., 2005). Undermining assumes that exhausted employees make
mistakes costing them even more effort and time to fix, which in turn leads to more
exhaustion and consequently even more mistakes (Linden et al., 2005). In addition, the
differences of experience level could reflect a maturation effect similar to those found in
personality traits and depression (Costa & McCrae, 1988; Debast et al., 2014; Prenoveau et
al., 2011), where it is assumed that these constructs display their biggest changes until
individuals reach early adulthood, when change slows down significantly. Although our
current sample is beyond early adulthood, we would still expect greater changes in less
experienced principals.
Other constructs with distinct components are emotions which also consist of state and
trait-like components (Zelenski & Larsen, 2000). For example trait anxiety is dispositional,
enduring from birth or early childhood, and presents across settings, while state anxiety is
more simply conceptualised around a highly anxiety provoking stimulus (e.g., Spielberger,
1972). The nature of emotional exhaustion is, however, very different than that of anxiety in

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
29
that even when it is considered trait-like, it is likely that individuals are not born with
emotional exhaustion or carry it from a young age, but rather that trait-like emotional
exhaustion is the result of a burnout specific constellation of chronic demands and resources.
Similarly, for state-like emotional exhaustion, which cannot be provoked through one single
stimulus, as can for instance state-anxiety by an anxiety-inducing stimulus, but can only occur
in burnout-specific situations (i.e., the interaction of a demanding situation and a lack of either
job related or personal resources to overcome these (Bakker & Demerouti, 2014; Hobfoll,
2001). These findings show the importance of understanding emotional exhaustion as a
reaction to an imbalance of demands and resources as in JD-theory (Bakker and Demerouti
2017), but also as a self-reinforcing process as in COR (Hobfoll 2001). This means there
might be a threshold, after which the continuing experience of emotional exhaustion could
lead to a manifestation of the experience of emotional exhaustion.
However, while a temporal aspect most likely plays a large role in the development
from state-like to trait-like emotional exhaustion, there might be individuals that are more
prone to develop or end up in a negative burnout spiral, supported by our results showing that
females seemed to display a larger trait-like emotional exhaustion. Concurrently, females
generally report higher levels of associated illbeing, including burnout and emotional
exhaustion compared to men in the general population (Purvanova & Muros, 2010), and
specifically in school principals (Dadaczynski & Paulus, 2015). Taken together this could be
translated into a higher incidence rate of chronic burnout, or a stronger disposition for females
to experience burnout. Assuming that job related demands of male and female principals
should be similar (Dicke et al. 2018; Darmody and Smyth 2016), a possible explanation for
the different experience of emotional exhaustion could be factors outside of work such as
family related additional demands, less access to support, or less effective coping mechanisms
(for an overview see Purvanova and Muros 2010). According to JD-R, all of these

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
30
disadvantages, that is, additional demands, for females would result in more opportunities for
an imbalance of demands and resources. This would in turn lead to an additional depletion of
resources which then accelerates the transition of experiencing state-like emotional
exhaustion to experiencing enduring and trait-like exhaustion (Bakker and Demerouti 2014).
Likewise, for example, personality characteristics in teachers (Cano-García et al., 2005;
Langelaan et al., 2006) and affect in teachers (Thoresen et al., 2003), and personality traits in
general (Alarcon et al., 2009) have been repeatedly shown to be associated with emotional
exhaustion. Such personal dispositions to developing burnout would be of particular
importance for developing, preventing and treating emotional exhaustion, as will be discussed
in the following section.
Practical Implications to Prevent and Deal with a Multi-Componential Emotional
Exhaustion
Given the large and similar magnitude of variance components associated with the
stable trait, enduring autoregressive trait, and occasion specific state component it is likely
that interventions that target all three are necessary. This includes event and time critical
resourcing to offset spikes in demands, as well as clinical interventions to improve coping
styles and protective resources (Bakker & Demerouti, 2014; Hobfoll, 2001), both ideally
targeted towards individual needs of educators (for an overview see Dicke, Elling, et al.,
2015; Vercambre et al., 2009). For inexperienced principals, emotional exhaustion was
mainly based on malleable components. This suggests the critical importance of early
mentoring, training, and support in order to avoid acute burnout transitioning to becoming
chronic (Dicke et al., 2014, 2016; Dicke, Elling, et al., 2015; Dicke, Parker, et al., 2015). This
is because inexperienced principal burnout symptoms appear to be far from set and thus most
amenable to change. While such early support will ideally function as a preventive measure to
avoid developing a more chronic trait-like manifestation of experiencing emotional

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
31
exhaustion, as seen more in experienced principals, additional interventions will still be
necessary for those with existing stable trait-like (and enduring autoregressive) emotional
exhaustion. Similarly to the above, such late interventions should aim to break the negative
burnout cycle by increasing the depleted resources and reducing the draining demands
(Bakker & Costa, 2014; Bakker & Demerouti, 2014; Hobfoll, 2001) to resolve the burnout
inducing situation. Thus, the onus of being mentally healthy should not be placed entirely on
principals. It is key that systemic changes to lessen the burden and demands of principals are
undertaken. Nevertheless, such rather slow-moving, politically-intertwined systemic changes
need to be complemented by more immediate effects through interventions aimed at the
resource and resilience building capacity of individuals, i.e. principals.
Among such measures to target ongoing, that is chronic, emotional exhaustion,
teaching employees how to recover and detach from work has proven to be helpful (Bakker &
Costa, 2014). Moreover, a recent meta-analyses of burnout interventions has shown that
particularly for emotional exhaustion, interventions that focus on relaxation techniques,
mindfulness, improving job related skills, and cognitive behavioral interventions (CBT) are
effective (for teachers Iancu et al., 2017; in general Maricuţoiu et al., 2016). Assuming that
there might be individuals that have a higher disposition to develop burnout would
furthermore suggest researching the histories of people who choose to take up the role as there
may be person-environment fit issues that only emerge over time (Burisch, 2002; Friedman,
2002).
In our results, the enduring autoregressive component of emotional exhaustion was of
a similar size to the other components and statistically significant. From an implication
perspective this means that policy makers and managers should provide support to principals,
not only immediately following an event that may result in a spike in burnout, but also over a
longer period given the possibility of lingering effects once the event has passed.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
32
Strengths, Limitations, and Implications for Future Research
The present study was based on a large, representative, and longitudinal sample of
Australian school principals. This is a strength for deriving conclusions for school principals,
but also a limitation in the sense that our results cannot be easily generalized. That said, the
school principals occupation is a very varied one (Dadaczynski & Paulus, 2015; Liebowitz &
Porter, 2019), which includes many tasks of administrative, financial, managing, leadership-
related, teaching-orientated nature.Thus, results found in this sample, might apply to many
other occupations. While the specific emotional exhaustion items used in the present study
themselves are worded “more generally”, the complete survey is framed to assess principal
health and wellbeing in the workplace and includes a multitude of references to being work-
related. Given this information, it is unlikely, but still possible that we assessed general
emotional exhaustion and not specifically work-related emotional exhaustion. Further, we
only focussed on one dimension of burnout, that is emotional exhaustion, which is a very
common approach, particularly when working with teacher (related) occupations (Dicke,
Stebner, et al., 2018; Iancu et al., 2017). Nevertheless, future research should compare the
decomposition of emotional exhaustion and other burnout components over different
occupations. An interesting focus in this regard would be to focus on the temporal sequence
of the burnout dimensions, thereby investigating the relation of acute and chronic burnout
with personal accomplishment and depersonalization, as research has suggested emotional
exhaustion precedes the development of these latter dimensions. While our results found
heterogeneity in the size of the variance components of emotional exhaustion to be linked to
individual characteristics (gender and experience) and not the job/school characteristics
(school sector or level), including other such variables relevant to JD-R might show a very
different pattern and should be considered in future research. The inclusion of other important
constructs as outcomes could also shed light on the validity of our assigned components,

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
33
research questions here could address if these outcomes are predicted differently by the
different components (Merz and Roesch 2011). Moreover, qualitative research following up
on the present study to understand the daily lived experience of principals would be
important. This research would be able to identify the more personal and individual
differences in the principal's development of emotional exhaustion. As in most burnout
studies (Bakker & Costa, 2014), our study made use of a convenience sample, where the most
affected (exhausted) principals most likely had dropped out. However additional analyses (see
Supplemental Material) did not show any meaningful or systematic relations between non-
response and levels of emotional exhaustion.
While the STARTS model has many strengths, such as the inclusion of three variance
components, controlling for measurement error, and flexibility in its setup when using
structural equation modelling, a limitation in its application is that it is quite complex and
prone to fail convergence (Kenny & Zautra, 2001). While all of our models converged, we
were not able to add even more complexity such as testing a model without any assumptions
of stationarity (i.e., allowing the size of absolute variance in addition to the distribution of
components to vary over time (Donnellan et al., 2012). With regard to the time intervals
chosen to measure emotional exhaustion we used an annual measurement. This choice was
foremost driven by theoretical (see Method section), but also practical considerations. School
principals already experience high levels of demands at work and long work hours (Darmody
& Smyth, 2016; Dicke et al., 2019; Dicke, Marsh, et al., 2018; Riley et al., 2019) It is, thus,
very important to regularly survey their levels of occupational work. But this work is also
quite difficult. Maintaining the high response levels, trust, and commitment of such samples is
only possible due to continuous effort put into the relationship, ongoing communication with
participants, and targeted marketing efforts. These measures are quite costly with regard to
resources. Further, it is crucial not to provoke research and survey fatigue within these

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
34
samples.
The point of the model is not to capture the person's state every day of the year, but
rather at the moment the survey is filled out with the assumption that principals will vary in
what is affecting them now. Further, research may want to consider the influence of survey
instructions related to time-frames (e.g., over the last week, today, right now, generally) on
the proportion of variance associated with state, enduring autoregressive, and trait. Another
interesting extension of the model could be to include other time-invariant and time-varying
covariates to address research questions that target potential predictions of the trait component
or state and autoregressive component respectively.
Our results have important implications for future research. The almost balanced
distribution of all three components found in our study highlights the critical importance of
distinguishing between- and within-person variance in burnout in applied research. While
cross-lagged models have been popular in burnout research (Bakker & Demerouti, 2007), too
few have taken advantage of recent developments in statistical analysis that aim to more
clearly focus on modelling and predicting between and within person change (e.g., bivariate
STARTS, Random Intercept-Cross Lag Panel Model; but see Aldrup et al., 2017; Bakker &
Costa, 2014).
Conclusion
The present study provided important insights into a) the development of emotional
exhaustion of a still understudied but high risk occupational group of school principals, b) the
decomposition of emotional exhaustion into state, enduring autoregressive, and trait-like
variance, and c) possible heterogeneity in these components. We found evidence for all three
components, with differences attributable to individual characteristics (experience and
gender) rather than school characteristics (school type and level). These results are important
for furthering the theoretical definition of emotional exhaustion and thus, burnout, but more

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
35
importantly for practical applications including prevention and treatment of high levels of
emotional exhaustion. Overall, we conclude that the responsibility for principals’ wellbeing
lies within the system and its policies as much as with the principals themselves. Moreover, it
is of critical importance to nip the development of chronic emotional exhaustion in the bud by
offering preventive measures to principals in burnout provoking situations. Thus, measures to
tackle emotional exhaustion should ideally include elements that tap both the
situational/contextual and individual factors that cause emotional exhaustion in school
principals. Furthermore, measures need to be individually targeted, meaning more preventive
for principals with more state-like emotional exhaustion, more interventive for principals with
more trait-like emotional exhaustion.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
36
References
Alarcon, G., Eschleman, K. J., & Bowling, N. A. (2009). Relationships between
personality variables and burnout: A meta-analysis. Work & Stress, 23(3), 244–263.
https://doi.org/10.1080/02678370903282600
Aldrup, K., Klusmann, U., & Lüdtke, O. (2017). Does basic need satisfaction mediate
the link between stress exposure and well-being? A diary study among beginning
teachers. Learning and Instruction, 50, 21–30.
https://doi.org/10.1016/j.learninstruc.2016.11.005
Alessandri, G., Zuffianò, A., Vecchione, M., Donnellan, B. M., & Tisak, J. (2016).
Evaluating the temporal structure and correlates of daily self-esteem using a trait state
error framework (TSE). Self and Identity, 15(4), 394–412.
https://doi.org/10.1080/15298868.2015.1137223
Allen, B. P., & Potkay, C. R. (1981). On the arbitrary distinction between states and
traits. Journal of Personality and Social Psychology, 41(5), 916–928.
https://doi.org/10.1037/0022-3514.41.5.916
Antoniou, A., Polychroni, F., & Walters, B. (2000). Sources of stress and professional
burnout of teachers of special educational needs in Greece. International Conference of
Special Education.
Anusic, I., Lucas, R. E., & Donnellan, M. B. (2012). Dependability of personality, life
satisfaction, and affect in short-term longitudinal data. Journal of Personality, 80(1), 33–
58. https://doi.org/10.1111/j.1467-6494.2011.00714.x
Anusic, I., & Schimmack, U. (2016). Stability and change of personality traits, self-
esteem, and well-being: Introducing the meta-analytic stability and change model of
retest correlations. Journal of Personality and Social Psychology, 110(5), 766–781.
https://doi.org/10.1037/pspp0000066

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
37
Asparouhov, T., & Muthén, B. (2010). Bayesian analysis using Mplus: Technical
implementation. Mplus Technical Report. http://statmodel. com/download/Bayes3.pdf
Bakker, A. B., & Costa, P. L. (2014). Chronic job burnout and daily functioning: A
theoretical analysis. Burnout Research, 1(3), 112–119.
https://doi.org/10.1016/j.burn.2014.04.003
Bakker, A. B., & Demerouti, E. (2007). The Job Demands‐Resources model: state of the
art. Journal of Managerial Psychology, 22(3), 309–328.
https://doi.org/10.1108/02683940710733115
Bakker, A. B., & Demerouti, E. (2014). Job Demands-Resources Theory. In C. L.
Cooper (Ed.), Wellbeing: A complete reference guide (pp. 1–28). John Wiley & Sons,
Ltd. https://doi.org/10.1002/9781118539415.wbwell019
Bakker, A. B., Demerouti, E., & Sanz-Vergel, A. I. (2014). Burnout and Work
Engagement: The JD–R Approach. Annual Review of Organizational Psychology and
Organizational Behavior, 1(1), 389–411. https://doi.org/10.1146/annurev-orgpsych-
031413-091235
Beausaert, S., Froehlich, D. E., Devos, C., & Riley, P. (2016). Effects of support on
stress and burnout in school principals. Educational Research, 58(4), 347–365.
https://doi.org/10.1080/00131881.2016.1220810
Brenninkmeyer, V., Van Yperen, N. W., & Buunk, B. P. (2001). Burnout and depression
are not identical twins: is decline of superiority a distinguishing feature? Personality and
Individual Differences, 30(5), 873–880. https://doi.org/10.1016/S0191-8869(00)00079-9
Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit.
Sociological Methods & Research, 21(2), 230–258.
https://doi.org/10.1177/0049124192021002005
Burisch, M. (2002). A longitudinal study of burnout: The relative importance of

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
38
dispositions and experiences. Work & Stress, 16(1), 1–17.
https://doi.org/10.1080/02678370110112506
Cano-García, F. J., Padilla-Muñoz, E. M., & Carrasco-Ortiz, M. Á. (2005). Personality
and contextual variables in teacher burnout. Personality and Individual Differences,
38(4), 929–940. https://doi.org/10.1016/j.paid.2004.06.018
Cattell, R. B. (1966). Patterns of change: Measurement in relation to state dimension,
trait change, lability, and process concepts. Handbook of Multivariate Experimental
Psychology, 355–402.
Costa, P. T., & McCrae, R. R. (1988). Personality in adulthood: A six-year longitudinal
study of self-reports and spouse ratings on the NEO Personality Inventory. Journal of
Personality and Social Psychology, 54(5), 853–863. https://doi.org/10.1037/0022-
3514.54.5.853
Cropanzano, R., Rupp, D. E., & Byrne, Z. S. (2003). The relationship of emotional
exhaustion to work attitudes, job performance, and organizational citizenship behaviors.
The Journal of Applied Psychology, 88(1), 160–169. https://doi.org/10.1037/0021-
9010.88.1.160
Dadaczynski, K., & Paulus, P. (2015). Healthy principals – healthy schools? A neglected
perspective to school health promotion. In V. Simovska & P. Mannix McNamara (Eds.),
Schools for health and sustainability (pp. 253–273). Springer Netherlands.
https://doi.org/10.1007/978-94-017-9171-7_12
Darmody, M., & Smyth, E. (2016). Primary school principals ’ job satisfaction and
occupational stress. International Journal of Educational Management, 30(1), 115–128.
https://doi.org/10.1108/IJEM-12-2014-0162
Day, D. V. (2011). Leadership development. In A. Bryman, D. Collinson, K. Grint, B.
Jackson, & M. Uhil-Bien (Eds.), The SAGE handbook of Leadership (pp. 37–50). Sage.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
39
Debast, I., van Alphen, S. P. J. (Bas), Rossi, G., Tummers, J. H. A., Bolwerk, N.,
Derksen, J. J. L., & Rosowsky, E. (2014). Personality Traits and Personality Disorders in
Late Middle and Old Age: Do They Remain Stable? A Literature Review. Clinical
Gerontologist, 37(3), 253–271. https://doi.org/10.1080/07317115.2014.885917
Dewa, C. S., Dermer, S. W., Chau, N., Lowrey, S., Mawson, S., & Bell, J. (2009).
Examination of factors associated with the mental health status of principals. Work
(Reading, Mass.), 33(4), 439–448. https://doi.org/10.3233/WOR-2009-0892
Dicke, T., Elling, J., Schmeck, A., & Leutner, D. (2015). Reducing reality shock: The
effects of classroom management skills training on beginning teachers. Teaching and
Teacher Education, 48(48), 1–12. https://doi.org/10.1016/j.tate.2015.01.013
Dicke, T., Holzberger, D., Kunina-Habenicht, O., Linninger, C., & Schulze-Stocker, F.
(2016). „Doppelter Praxisschock“ auf dem Weg ins Lehramt? Verlauf und potenzielle
Einflussfaktoren emotionaler Erschöpfung während des Vorbereitungsdienstes und nach
dem Berufseintritt. Psychologie in Erziehung Und Unterricht, 63(4), 244.
https://doi.org/10.2378/peu2016.art20d
Dicke, T., Marsh, H. W., Parker, P. D., Guo, J., Riley, P., & Waldeyer, J. (2019). Job
satisfaction of teachers and their principals in relation to climate and student
achievement. Journal of Educational Psychology, 112(5), 1061.
https://doi.org/10.1037/edu0000409
Dicke, T., Marsh, H. W., Riley, P., Parker, P. D., Guo, J., & Horwood, M. (2018).
Validating the Copenhagen Psychosocial Questionnaire (COPSOQ-II) Using Set-ESEM:
Identifying Psychosocial Risk Factors in a Sample of School Principals. Frontiers in
Psychology, 9, 584. https://doi.org/10.3389/fpsyg.2018.00584
Dicke, T., Parker, P. D., Holzberger, D., Kunina-Habenicht, O., Kunter, M., & Leutner,
D. (2015). Beginning teachers’ efficacy and emotional exhaustion: Latent changes,

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
40
reciprocity, and the influence of professional knowledge. Contemporary Educational
Psychology, 41, 62–72. https://doi.org/10.1016/j.cedpsych.2014.11.003
Dicke, T., Parker, P. D., Marsh, H. W., Kunter, M., Schmeck, A., & Leutner, D. (2014).
Self-efficacy in classroom management, classroom disturbances, and emotional
exhaustion: A moderated mediation analysis of teacher candidates. Journal of
Educational Psychology, 106(2), 569–583. https://doi.org/10.1037/a0035504
Dicke, T., Stebner, F., Linninger, C., Kunter, M., & Leutner, D. (2018). A longitudinal
study of teachers’ occupational well-being: Applying the job demands-resources model.
Journal of Occupational Health Psychology, 23(2), 262–277.
https://doi.org/10.1037/ocp0000070
Donnellan, M. B., Kenny, D. A., Trzesniewski, K. H., Lucas, R. E., & Conger, R. D.
(2012). Using Trait-State Models to Evaluate the Longitudinal Consistency of Global
Self-Esteem From Adolescence to Adulthood. Journal of Research in Personality, 46(6),
634–645. https://doi.org/10.1016/j.jrp.2012.07.005
Enders, C. K. (2010). Applied Missing Data Analysis (illustrated). Guilford Press.
Eysenck, H. J. (1983). Cicero and the state-trait theory of anxiety: Another case of
delayed recognition. American Psychologist, 38(1), 114–115.
https://doi.org/10.1037/0003-066X.38.1.114
Fridhandler, B. M. (1986). Conceptual note on state, trait, and the statetrait distinction.
Journal of Personality and Social Psychology, 50(1), 169–174.
https://doi.org/10.1037/0022-3514.50.1.169
Friedman, I. A. (2002). Burnout in school principals: role related antecedents. Springer
Science and Business Media LLC. https://doi.org/10.1023/a:1016321210858
Gelman, A., & Rubin, D. B. (1992). Inference from Iterative Simulation Using Multiple
Sequences. Statistical Science, 7(4), 457–472. https://doi.org/10.1214/ss/1177011136

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
41
Goldring, R. (2014). Principal Attrition and Mobility: Results From the 201213
Principal Follow-up Survey (S. Taie, trans.). National Center for Education Statistics.
Goldring, R. (2018). Principal Attrition and Mobility: Results From the 201617 Principal
Follow-up Survey (NCES 2018-066) (S. Taie, trans.). National Center for Education
Statistics.
Golembiewski, R., & Stevenson, J. (1985). Stress in Organizations: Toward A Phase
Model of Burnout. Praeger.
Golembiewski, R. T., & Munzenrider, R. F. (1988). Phases of Burnout: Developments in
Concepts and Applications. Praeger Publishers.
Green, R., Malcolm, S., Greenwood, K., Small, M., & Murphy, G. (2001). A survey of
the health of Victorian primary school principals. International Journal of Educational
Management, 15(1), 23–30. https://doi.org/10.1108/09513540110366114
Grissom, J. A., Loeb, S., & Mitani, H. (2015). Principal time management skills. Journal
of Educational Administration, 53(6), 773–793. https://doi.org/10.1108/JEA-09-2014-
0117
Hamaker, E. L., Kuiper, R. M., & Grasman, R. P. P. P. (2015). A critique of the cross-
lagged panel model. Psychological Methods, 20(1), 102–116.
https://doi.org/10.1037/a0038889
Hamaker, E. L., Nesselroade, J. R., & Molenaar, P. C. M. (2007). The integrated trait–
state model. Journal of Research in Personality, 41(2), 295–315.
https://doi.org/10.1016/j.jrp.2006.04.003
Hanselman, P., Grigg, J., K. Bruch, S., & Gamoran, A. (2016). The consequences of
principal and teacher turnover for school social resources. In G. Kao & H. Park (Eds.),
Family environments, school resources, and educational outcomes (Vol. 19, pp. 49–89).
Emerald Group Publishing Limited. https://doi.org/10.1108/S1479-

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
42
353920150000019004
Hertzog, C., & Nesselroade, J. R. (1987). Beyond Autoregressive Models: Some
Implications of the Trait-State Distinction for the Structural Modeling of Developmental
Change. Child Development, 58(1), 93. https://doi.org/10.2307/1130294
Hobfoll, S. E. (2001). The Influence of Culture, Community, and the Nested-Self in the
Stress Process: Advancing Conservation of Resources Theory. Applied Psychology,
50(3), 337–421. https://doi.org/10.1111/1464-0597.00062
Huang, Y.-H., Du, P.-I., Chen, C.-H., Yang, C.-A., & Huang, I.-C. (2011). Mediating
effects of emotional exhaustion on the relationship between job demand–control model
and mental health. Stress and Health : Journal of the International Society for the
Investigation of Stress, 27(2), e94-e109. https://doi.org/10.1002/smi.1340
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A
Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118
Iancu, A. E., Rusu, A., Măroiu, C., Păcurar, R., & Maricuțoiu, L. P. (2017). The
Effectiveness of Interventions Aimed at Reducing Teacher Burnout: a Meta-Analysis.
Educational Psychology Review, 30(2), 1–24. https://doi.org/10.1007/s10648-017-9420-
8
Jansen, M., Lüdtke, O., & Robitzsch, A. (2020). Disentangling different sources of
stability and change in students’ academic self-concepts: An integrative data analysis
using the STARTS model. Journal of Educational Psychology, 112(8), 1614–1631.
https://doi.org/10.1037/edu0000448
Jelicić, H., Phelps, E., & Lerner, R. M. (2009). Use of missing data methods in
longitudinal studies: the persistence of bad practices in developmental psychology.
Developmental Psychology, 45(4), 1195–1199. https://doi.org/10.1037/a0015665

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
43
Kenny, David A., & Zautra, A. (2001). Trait–state models for longitudinal data. In L. M.
Collins & A. G. Sayer (Eds.), New methods for the analysis of change. (pp. 243–263).
American Psychological Association. https://doi.org/10.1037/10409-008
Kenny, D A, & Zautra, A. (1995). The trait-state-error model for multiwave data.
Journal of Consulting and Clinical Psychology, 63(1), 52–59.
https://doi.org/10.1037//0022-006x.63.1.52
Klassen, R. M., & Chiu, M. M. (2010). Effects on teachers’ self-efficacy and job
satisfaction: Teacher gender, years of experience, and job stress. Journal of Educational
Psychology, 102(3), 741–756. https://doi.org/10.1037/a0019237
Klusmann, U., Aldrup, K., Schmidt, J., & Lüdtke, O. (2020). Is emotional exhaustion
only the result of work experiences? A diary study on daily hassles and uplifts in
different life domains. Anxiety, Stress, and Coping, 1–18.
https://doi.org/10.1080/10615806.2020.1845430
Klusmann, U., Kunter, M., Trautwein, U., Lüdtke, O., & Baumert, J. (2008).
Engagement and emotional exhaustion in teachers: does the school context make a
difference? Applied Psychology, 57(s1), 127–151. https://doi.org/10.1111/j.1464-
0597.2008.00358.x
Klusmann, U., Richter, D., & Lüdtke, O. (2016). Teachers’ emotional exhaustion is
negatively related to students’ achievement: Evidence from a large-scale assessment
study. Journal of Educational Psychology, 108(8), 1193–1203.
https://doi.org/10.1037/edu0000125
Kristensen, T. S., Hannerz, H., Høgh, A., & Borg, V. B. (2005). The Copenhagen
Psychosocial Questionnaire-a tool for the assessment and improvement of the
psychosocial work environment on JSTOR. Scandinavian Journal of Work, Environment
& Health, 31(6), 438–449.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
44
Langelaan, S., Bakker, A. B., van Doornen, L. J. P., & Schaufeli, W. B. (2006). Burnout
and work engagement: Do individual differences make a difference? Personality and
Individual Differences, 40(3), 521–532. https://doi.org/10.1016/j.paid.2005.07.009
Lee, R. T., & Ashforth, B. E. (1993). A further examination of managerial burnout:
Toward an integrated model. Journal of Organizational Behavior, 14(1), 3–20.
https://doi.org/10.1002/job.4030140103
Leithwood, K., & Seashore-Louis, K. (2011). Linking Leadership To Student Learning
(1st ed., p. 320). Jossey-bass.
Liebowitz, D. D., & Porter, L. (2019). The Effect of Principal Behaviors on Student,
Teacher, and School Outcomes: A Systematic Review and Meta-Analysis of the
Empirical Literature. Review of Educational Research, 89(5), 785–827.
https://doi.org/10.3102/0034654319866133
Linden, D. V. D., Keijsers, G. P. J., Eling, P., & Schaijk, R. V. (2005). Work stress and
attentional difficulties: An initial study on burnout and cognitive failures. Work & Stress,
19(1), 23–36. https://doi.org/10.1080/02678370500065275
Maricuţoiu, L. P., Sava, F. A., & Butta, O. (2016). The effectiveness of controlled
interventions on employees’ burnout: A meta-analysis. Journal of Occupational and
Organizational Psychology, 89(1), 1–27. https://doi.org/10.1111/joop.12099
Marsh, H. W., & Grayson, D. (1994). Longitudinal stability of latent means and
individual differences: A unified approach. Structural Equation Modeling: A
Multidisciplinary Journal, 1(4), 317–359. https://doi.org/10.1080/10705519409539984
Marsh, H. W., Hau, K.-T., & Wen, Z. (2004). In Search of Golden Rules: Comment on
Hypothesis-Testing Approaches to Setting Cutoff Values for Fit Indexes and Dangers in
Overgeneralizing Hu and Bentler’s (1999) Findings. Structural Equation Modeling: A
Multidisciplinary Journal, 11(3), 320–341.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
45
https://doi.org/10.1207/s15328007sem1103_2
Maslach, Christina, Jackson, S. E., & Leiter, M. P. (1996). The Maslach Burnout
Inventory (3rd ed.). Consulting Psychologists Press.
Maslach, Christina, Schaufeli, W. B., & Leiter, M. P. (2001). Job Burnout. Annual
Review of Psychology, 52, 397–422. https://doi.org/10.1146/annurev.psych.52.1.397
Maslach, Christine, Jackson, S. E., & Leiter, M. P. (1997). Maslach Burnout Inventory:
Third edition. In C. P. Zalaquett & R. J. Wood (Eds.), Evaluating stress: A book of
resources (pp. 191–218). Scarecrow Education.
Maslach, C, & Leiter, M. P. (2016). Understanding the burnout experience: Recent
research and its implications for psychiatry. World Psychiatry : Official Journal of the
World Psychiatric Association (WPA), 15(2), 103–111.
https://doi.org/10.1002/wps.20311
Maxwell, A., & Riley, P. (2017). Emotional demands, emotional labour and occupational
outcomes in school principals. Educational Management Administration & Leadership,
45(3), 484–502. https://doi.org/10.1177/1741143215607878
Mcdonald, R. P. (1999). Test Theory: A Unified Treatment (1st ed., p. 498). L. Erlbaum
Associates.
Milligan-Saville, J., Choi, I., Deady, M., Scott, P., Tan, L., Calvo, R. A., Bryant, R. A.,
Glozier, N., & Harvey, S. B. (2018). The impact of trauma exposure on the development
of PTSD and psychological distress in a volunteer fire service. Psychiatry Research, 270,
1110–1115. https://doi.org/10.1016/j.psychres.2018.06.058
Mund, M., Lüdtke, O., & Neyer, F. J. (2020). Owner of a lonely heart: The stability of
loneliness across the life span. Journal of Personality and Social Psychology, 119(2),
497–516. https://doi.org/10.1037/pspp0000262
Muthén, L. K., & Muthén, B. O. (2012). Mplus User’s Guide. Seventh Edition. (Version

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
46
7) [Computer manual]. Muthén & Muthén.
Newman, D. A. (2014). Missing Data: Five Practical Guidelines. Organizational
Research Methods, 17(4), 372–411. https://doi.org/10.1177/1094428114548590
Prenoveau, J. M., Craske, M. G., Zinbarg, R. E., Mineka, S., Rose, R. D., & Griffith, J.
W. (2011). Are anxiety and depression just as stable as personality during late
adolescence? Results from a three-year longitudinal latent variable study. Journal of
Abnormal Psychology, 120(4), 832–843. https://doi.org/10.1037/a0023939
Purvanova, R. K., & Muros, J. P. (2010). Gender differences in burnout: A meta-
analysis. Journal of Vocational Behavior, 77(2), 168–185.
https://doi.org/10.1016/j.jvb.2010.04.006
Riley, P. (2014). The Australian principal occupational health,safety & well-being
survey: 2011-2014 data. Australian Catholic University.
Riley, P. (2015). The Australian principal occupational health, safety & well-being
survey: 2015 data. Australian Catholic University.
Riley, P. (2017). The Australian principal occupational health, safety & well-being
survey: 2017 data. Australian Catholic University.
Riley, P. (2018). The Australian principal occupational health, safety & well-being
survey: 2018 data. Australian Catholic University.
Riley, P., See, S. M., & Dicke, T. (2019). The Australian Principal Occupational Health,
Safety, and Wellbeing Survey. . Institute for Positive Psychology and Education,
Australian Catholic University.
Schafer, J. L., & Graham, J. W. (2002). Missing data: our view of the state of the art.
Psychological Methods, 7(2), 147–177. https://doi.org/10.1037/1082-989X.7.2.147
Schaufeli, W. (1998). The Burnout Companion To Study And Practice (issues In
Occupational Health Series) (1st ed., p. 224). Routledge.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
47
Sebastian, J., & Allensworth, E. (2012). The Influence of Principal Leadership on
Classroom Instruction and Student Learning: A Study of Mediated Pathways to
Learning. Educational Administration Quarterly, 48(4), 626–663.
https://doi.org/10.1177/0013161X11436273
Shen, B., McCaughtry, N., Martin, J., Garn, A., Kulik, N., & Fahlman, M. (2015). The
relationship between teacher burnout and student motivation. The British Journal of
Educational Psychology, 85(4), 519–532. https://doi.org/10.1111/bjep.12089
Simbula, S. (2010). Daily fluctuations in teachers’ well-being: a diary study using the
Job Demands-Resources model. Anxiety, Stress, and Coping, 23(5), 563–584.
https://doi.org/10.1080/10615801003728273
Sonnentag, S. (2005). Burnout research: Adding an off-work and day-level perspective
The views expressed in Work & Stress Commentaries are those of the author(s), and do
not necessarily represent those of any other person or organization, or of the journal.
Work & Stress, 19(3), 271–275. https://doi.org/10.1080/02678370500386473
Spielberger, C. D. (1972). Anxiety as an Emotional State. In Anxiety. Current Trends in
Theory and Research (pp. 23–49). Elsevier. https://doi.org/10.1016/B978-0-12-657401-
2.50009-5
Steyer, R., Geiser, C., & Fiege, C. (2012). Latent state-trait models. In H. Cooper, P. M.
Camic, D. L. Long, A. T. Panter, D. Rindskopf, & K. J. Sher (Eds.), APA handbook of
research methods in psychology, Vol 3: Data analysis and research publication. (pp.
291–308). American Psychological Association. https://doi.org/10.1037/13621-014
Steyer, R., Mayer, A., Geiser, C., & Cole, D. A. (2015). A theory of states and traits--
revised. Annual Review of Clinical Psychology, 11, 71–98.
https://doi.org/10.1146/annurev-clinpsy-032813-153719
The Effect of Principal Leadership on Student Achievement: A Multivariate Meta-

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
48
Analysis. (2020). Dissertations.
Thoresen, C. J., Kaplan, S. A., Barsky, A. P., Warren, C. R., & de Chermont, K. (2003).
The affective underpinnings of job perceptions and attitudes: a meta-analytic review and
integration. Psychological Bulletin, 129(6), 914–945. https://doi.org/10.1037/0033-
2909.129.6.914
Vercambre, M.-N., Brosselin, P., Gilbert, F., Nerrière, E., & Kovess-Masféty, V. (2009).
Individual and contextual covariates of burnout: a cross-sectional nationwide study of
French teachers. BMC Public Health, 9, 333. https://doi.org/10.1186/1471-2458-9-333
Wagner, J., Lüdtke, O., & Trautwein, U. (2016). Self-esteem Is Mostly Stable Across
Young Adulthood: Evidence from Latent STARTS Models. Journal of Personality,
84(4), 523–535. https://doi.org/10.1111/jopy.12178
Weber, A., Weltle, D., & Lederer, P. (2005). Ill health and early retirement among
school principals in Bavaria. International Archives of Occupational and Environmental
Health, 78(4), 325–331. https://doi.org/10.1007/s00420-004-0555-9
Wells, C. M., & Klocko, B. A. (2018). Principal Well-Being and Resilience:
Mindfulness as a Means to That End. NASSP Bulletin, 102(2), 161–173.
https://doi.org/10.1177/0192636518777813
Wright, T. A., Cropanzano, R., & Meyer, D. G. (2003). State and trait correlates of job
performance: A tale of two perspectives. Journal of Business and Psychology, 18(3),
365–383. https://doi.org/10.1023/B:JOBU.0000016708.22925.72
Zelenski, J. M., & Larsen, R. J. (2000). The Distribution of Basic Emotions in Everyday
Life: A State and Trait Perspective from Experience Sampling Data. Journal of Research
in Personality, 34(2), 178–197. https://doi.org/10.1006/jrpe.1999.2275
Zinbarg, R. E., Yovel, I., Revelle, W., & McDonald, R. P. (2006). Estimating
Generalizability to a Latent Variable Common to All of a Scale’s Indicators: A

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
49
Comparison of Estimators for ωh. Applied Psychological Measurement, 30(2), 121–144.
https://doi.org/10.1177/0146621605278814

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
50
Figure 1
Basic STARTS Model
Note. STARTS (stable trait, autoregressive trait, and state) model based on multiple indicator
factors decomposing emotional exhaustion into a stable trait, an enduring autoregressive
component, an occasion specific state, and measurement error. AR = autoregressive trait. AR
Resid. = autoregressive residual. Correlated uniquenesses were included in the analyses but
are not displayed in this figure.

EMOTIONAL EXHAUSTION IN SCHOOL PRINCIPALS
51
Figure 2
Variance Components by Group
Note. Trait = stable trait. AR = enduring autoregressive component. State = occasion specific
state.